def test_rate_manager_crossval():
path = data_path / "fmt-jpk-fd_map-data-reference-points.jpk-force-map"
tdir, h5path = setuph5(path=path)
rmg = RateManager(h5path)
cv = rmg.get_cross_validation_score(regressor="Extra Trees",
training_set=None,
n_splits=2,
random_state=42)
assert np.all(cv == 0)
def test_rate_manager_get_ts_single():
tdir, h5path = setuph5()
rmg = RateManager(h5path)
x2, _ = rmg.get_training_set(which_type="binary")
assert np.all(x2 == 1)
x3, _ = rmg.get_training_set(which_type="continuous",
prefilter_binary=True)
x4, _ = rmg.get_training_set(remove_nans=True)
assert np.all(np.hstack((x2, x3)) == x4)
def test_rate_manager_get_ts_single():
tdir, h5path = setuph5()
rmg = RateManager(h5path)
X2, _ = rmg.get_training_set(which_type="binary")
assert np.all(X2 == 1)
X3, _ = rmg.get_training_set(which_type="continuous",
prefilter_binary=True)
X4, _ = rmg.get_training_set(remove_nans=True)
assert np.all(np.hstack((X2, X3)) == X4)
shutil.rmtree(tdir, ignore_errors=True)
def test_rate_manager_get_ts():
path = data_path / "fmt-jpk-fd_map-data-reference-points.jpk-force-map"
tdir, h5path = setuph5(path=path)
rmg = RateManager(h5path)
x2, _ = rmg.get_training_set(which_type="binary")
assert np.all(x2 == 1)
x3, _ = rmg.get_training_set(which_type="continuous",
prefilter_binary=True)
x4, _ = rmg.get_training_set(remove_nans=True)
assert np.all(np.hstack((x2, x3)) == x4)
def test_rate_manager_get_ts():
path = datadir / "map-data-reference-points.jpk-force-map"
tdir, h5path = setuph5(path=path)
rmg = RateManager(h5path)
X2, _ = rmg.get_training_set(which_type="binary")
assert np.all(X2 == 1)
X3, _ = rmg.get_training_set(which_type="continuous",
prefilter_binary=True)
X4, _ = rmg.get_training_set(remove_nans=True)
assert np.all(np.hstack((X2, X3)) == X4)
shutil.rmtree(tdir, ignore_errors=True)
def test_rate_manager_get_ts_bad():
path = data_path / "fmt-jpk-fd_map_bad_2013-05-27_2.jpk-force-map"
tdir, h5path = setuph5(path=path)
rmg = RateManager(h5path)
x2, _ = rmg.get_training_set(which_type="binary")
assert np.allclose(x2.flatten(), [1, 0, 1])
x3, _ = rmg.get_training_set(which_type="continuous",
prefilter_binary=True)
assert x3.size == 0
x4, _ = rmg.get_training_set(remove_nans=True)
assert x4.size == 0
def test_rate_manager_get_ts_bad():
path = datadir / "bad_map-data-2013.05.27-13.50.21.jpk-force-map"
tdir, h5path = setuph5(path=path)
rmg = RateManager(h5path)
X2, _ = rmg.get_training_set(which_type="binary")
assert np.allclose(X2.flatten(), [1, 0, 1])
X3, _ = rmg.get_training_set(which_type="continuous",
prefilter_binary=True)
assert X3.size == 0
X4, _ = rmg.get_training_set(remove_nans=True)
assert X4.size == 0
shutil.rmtree(tdir, ignore_errors=True)
def test_rate_manager_export():
tdir, h5path = setuph5()
rmg = RateManager(h5path)
rmg.export_training_set(tdir)
ss = rmg.samples[0]
feats = IndentationRater.get_feature_names()
for ff, si in zip(feats, ss):
fi = np.loadtxt(tdir / "train_{}.txt".format(ff))
# :.2e, because features are not stored with high accuracy
assert np.ndarray.item(fi) == float("{:.2e}".format(si))
def test_rate_manager_basic():
tdir, h5path, idnt = setuph5(ret_idnt=True)
rmg = RateManager(h5path)
# sanity checks
rr = rmg.ratings[0]
assert rr["name"] == "hans"
assert rr["rating"] == 5
# file name preserved
ds = rmg.datasets[0]
assert jpkfile.name in ds.path.name
# features are the same
idr = IndentationRater
ss = rmg.samples[0]
assert np.allclose(ss, idr.compute_features(idnt))
# rates
assert np.ndarray.item(rmg.get_rates(which="user")) == 5
# This will fail when the hyper-parameters for "Extra Trees" change
# or when new features are added.
assert np.allclose(
np.ndarray.item(rmg.get_rates(which="Extra Trees",
training_set="zef18")),
3.5492840783289035)
